Measurement errors in diffuse irradiance with non-Lambertian radiometers

Abstract

Experimental errors in measurements of solar irradiance resulting from the departure of angular response of real sensors from the Lambertian ideal are evaluated. The errors include polarization independent and dependent components. The polarization independent component is relatively independent of solar zenith angle and thus is substantially correctable by scaling. It was estimated to be as high as m 6.0% for some commonly used radiometers. The polarization dependent component for the multifilter rotating shadowband radiometer (MFRSR) was estimated to be between m 1.5% for the Sun at the zenith and +1.0% for the Sun at the horizon. The zero crossover for this error occurs around 55° solar zenith angle. These errors do not explain the discrepancy between measured and modelled diffuse shortwave irradiance, but in some instances they can reduce the difference.